Achieved anticoagulation vs prosthesis selection for mitral mechanical valve replacement: a population-based outcome study

Abstract

Background: Thromboembolic events (TEs) are frequent after mechanical mitral valve replacement (MVR), but their association to anticoagulation quality is unclear and has never been studied in a population-based setting with patients who have a complete anticoagulation record.

Methods: We compiled a complete record of all residents of Olmsted County, MN, who underwent mechanical MVR between 1981 and 2004, for all TE, bleeding episodes, and international normalized ratios (INRs) measured from prosthesis implantation.

Results: In the 112 residents (mean [+/- SD] age, 57 +/- 16 years; 60% female residents) who underwent mechanical MVR, 19,647 INR samples were obtained. While INR averaged 3.02 +/- 0.57, almost 40% of INRs were < 2 or > 4.5. Thirty-four TEs and 28 bleeding episodes occurred during a mean duration of 8.2 +/- 6.1 years of follow-up. There was no trend of association of INR (average, SD, growth variance rate, or intensity-specific incidence of events) with TE. Previous cardiac surgery (p = 0.014) and ball prosthesis (hazard ratio [HR], 2.92; 95% CI, 1.43 to 5.94; p = 0.003) independently determined TE. With MVR using a ball prosthesis, despite higher anticoagulation intensity (p = 0.002), the 8-year rate of freedom from TE was considerably lower (50 +/- 9% vs 81 +/- 5%, respectively; p < 0.0001). Compared with expected stroke rates in the population, stroke risk was elevated with non-ball prosthesis MVR (HR 2.6; 95% CI, 1.3 to 5.2; p = 0.007) but was considerable with ball prosthesis MVR (HR 11.7; 95% CI, 7.5 to 18.4; p < 0.0001). INR variability (SD) was higher with a higher mean INR value (p < 0.0001). INR variability (HR 2.485; 95% CI, 1.11 to 5.55; p = 0.027) and cancer history (p < 0.0001) independently determined bleeding rates.

Conclusion: This population-based comprehensive study of anticoagulation and TE post-MVR shows that, in these closely anticoagulated patients, anticoagulation intensity was highly variable and not associated with TE incidence post-MVR. Higher anticoagulation intensity is linked to higher variability and, thus, to bleeding. The MVR-ball prosthesis design is associated with higher TE rates notwithstanding higher anticoagulation intensity, and its use should be retired worldwide.

Figure 1

Anticoagulation variability in the entire patient population. A: distribution of all of INRs during long-term follow-up after mechanical MVR surgery. Note that almost 40% of INRs were outside of the extended therapeutic range. B: distribution of INR SD showing that most patients displayed a high variability of INR, reflecting a high variability of anticoagulation intensity.

Figure 2

The distribution of events (TE, left; bleeding, right) and of rates of events calculated as the ratio of events to time spent within 0.5 INR increments according to the Rosendaal method. Note that for an INR > 4.5, TE rates tend to be higher than for INRs within the 2 to 4.5 range; however, there are no significant differences among all ranges (p > 0.18). Note also that for an INR < 2, bleeding rates tend to be higher than INRs within the range of 2 to 4.5; however, there are no significant differences among all ranges (p > 0.48).

Figure 3

The distribution of INRs during the entire follow- up period in individual patients in the series. The solid symbols represent patients without TEs (no TE), and the open symbols show the patients with TEs. The INR SD is represented by the bar with each symbol. Note the considerable overlap between patients with and without TEs in terms of average INR and of SD.

Figure 4

The plot of INR-average (A and C) and INR SD (B and D) calculated up to the time of the event for TE (A and B) and for bleeding (C and D). For TE, there was no difference in INR-average or INR SD between those patients who had and had not experienced events. For bleeding, there was no difference in INR-average, but there was a higher INR SD in those patients who had experienced events. In patients who had not experienced events, the INR was calculated over the entire follow-up period.

Figure 5

Freedom from TE according to stratification by anticoagulation quality up to TEs (A and B) and by patients and prosthesis-related factors (C and D). Stratification was performed according to an INR SD ≥ 1 or < 1 (A), to INR-average ≥ 3.5 or < 3.5 (B), to a history of cardiac surgery (C), and to implantation of a ball prosthesis (D). Note the absence of a significant difference in TE rates between the strata defined by anticoagulation variability or intensity, whereas the large difference in TE rates is observed according to the implantation of a ball and non-ball mitral valve prosthesis.